|A randomized trial of two public health nurse follow-up programs after early obstetrical discharge. An examination of breastfeeding rates, maternal confidence and utilization and costs of health services.
|Steel O'Connor K O, Mowat D L, Scott H M, Carr P A, Dorland J L, Young Tai K F W
This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn.
The health technologies studied were two alternative forms of postpartum follow-up by public health nurses (PHN):
Telephone screen follow-up (TSFU), by which mothers were telephoned on the day following hospital discharge.
Home visit follow-up (HVFU), consisting of routine provision of two home visits within 10 days of discharge.
Under both forms of follow-up, other support services, primary medical care or community services were provided if needs were identified by either the mother or the PHN.
Type of intervention
Supportive care and primary prevention.
Economic study type
The study population comprised low-risk primiparas who delivered a singleton infant vaginally and were discharged within two days of the birth of their infants.
The setting was community care. The study was carried out in Ontario, Canada.
Dates to which data relate
The authors reported that the study period was between January 27, and January 31, 1999, but this may correspond to the period of recruitment of patients. Therefore, the effectiveness and resource utilisation data may have been collected between January and July 1999. The price year was January 1999.
Source of effectiveness data
The effectiveness data was derived from a single study.
Link between effectiveness and cost data
Resource quantities were obtained prospectively from one of the sub-sample populations used in the effectiveness analysis. As the authors reported, the full cost analysis was performed for subjects enrolled at site A and costs were imputed to resource utilisation of subjects enrolled at site B.
Power calculations were performed in the planning phase of the study in order to assure a certain power. A minimum sample size of 375 mothers per group was required in order to provide 80% power to detect a 10% difference in the breastfeeding rate between the interventions with 95% confidence. A minimum sample size of 336 mothers per group was required to provide 90% power to detect a difference in maternal confidence scores of 0.25 standard deviations with 95% confidence. Primiparas delivering vaginally a single infant during the study period in the hospitals considered at analysis, who resided in the areas served by the local Community Care Access Centre, and were able to understand English well enough (or French, in the case of patients recruited at site B), in order to give informed consent, were considered for the effectiveness analysis. The study sample comprised 733 mothers: 380 mothers were allocated to the TSFU group (with 184 enrolled at site A and 196 enrolled at site B) and 353 mothers were allocated to the HVFU group (with 174 enrolled at site A and 179 enrolled at site B). The authors did not report evidence that the study sample was appropriate for the study question.
This was a multi centred (two hospitals) randomised controlled trial. Sealed envelopes were used to allocate patients randomly to groups. Patients were followed-up during 28 days, except for those mothers that continued breastfeeding their infants during this period and they were followed for six months. 3.3% of mothers were lost to follow-up between intake and two weeks after birth, 1.8% were lost to follow-up between two and four weeks after birth, and 2.1% were lost between four weeks and six months. The authors stated that the research assistants who collected the data were blinded to the allocation of the mothers to study groups.
Analysis of effectiveness
Although the authors stated that the basis for the effectiveness analysis was intention-to-treat, some of the outcomes were assessed according to the available information at each assessment (i.e. at two weeks, four weeks, and six months). The primary outcomes were assessed not only for both types of postpartum follow-up (TSFU versus HVFU), but also by site. This was because the authors reported that the two sites considered at analysis were slightly different in terms of the training of the PHNs regarding the delivery of the follow-up care. The primary health outcomes assessed in the clinical study were: the overall participation rate of mothers at sites A and B; the number (and percentage) of mothers breastfeeding their infants at two weeks, four weeks, and six months after the birth of the infant; and the cumulative percentage of mothers breastfeeding their infants from discharge up to two weeks, four weeks and six months; the levels of maternal confidence related to infant care before discharge and two weeks after the birth of the infant; and the number (and percentage) of maternal reports of infant health problems, and the common odds ratios of TSFU patients when compared to HVFU patients (after demonstrating that the problems experienced by infants receiving the same type of follow-up at the two sites did not differ significantly).
The Maternal Confidence Scale, which used a five-point Likert scale, was used to assess the levels of maternal confidence (see Carty and Bradley, in 'Other Publications of Related Interest' below). Patient groups recruited at either site A or B were shown to be comparable at analysis in terms of gravidity and the sex of the infants, but not in terms of age, educational level of the mothers and gestational period. However, the authors did not report statistical analyses comparing TSFU and HVFU patients.
The participation rate among eligible women was 89.2% at site A and 80.0% at site B.
At site A, the number (and percentage) of mothers breastfeeding their infants were:
147/167 (87.0%) with TSFU versus 130/149 (87.2%) with HVFU at two weeks;
130/146 (89%) with TSFU versus 122/129 (94.6%) with HVFU at four weeks;
and 69/129 (53.5%) with TSFU versus 69/118 (58.5%) with HVFU at six months after the birth of the infant.
The cumulative percentage of women breastfeeding their infants were:
87% with TSFU versus 87.2% with HVFU at two weeks;
77.4% with TSFU versus 82.5% with HVFU at four weeks;
and 41.4% under TSFU versus 48.3% with HVFU at six months.
At site B, the number (and percentage) of mothers breastfeeding their infants were:
145/165 (87.9%) with TSFU versus 141/157 (89.8%) with HVFU at two weeks;
136/143 (95.1%) with TSFU versus 133/140 (94.6%) with HVFU at four weeks;
and 80/133 (60.2%) with TSFU versus 77/129 (59.7%) with HVFU at six months after the birth of the infant.
The cumulative percentage of women breastfeeding their infants were:
87.9% with TSFU versus 89.8% with HVFU at two weeks;
83.4% with TSFU versus 85.3% with HVFU at four weeks;
and 50.2% under TSFU versus 50.9% with HVFU at six months.
The confidence scores obtained in hospital (and the standard deviations, SD) were:
32.5 (SD: 4.6) for TSFU patients versus 32.6 (SD: 4.7) for HVFU patients at site A, and 31.0 (SD: 4.7) for TSFU patients versus 32.0 (SD: 4.8) for HVFU patients at site B.
The confidence scores (and SD) obtained at two weeks were:
34.4 (SD: 3.3) for TSFU versus 34.7 (SD: 3.4) for HVFU patients at site A and 34.0 (SD: 3.4) for TSFU patients versus 34.1 (SD: 3.5) for HVFU patients at site B.
As infants at the two sites experienced similar health problems, common odds ratios comparing TSFU and HVFU could be calculated.
These odds ratios did not differ significantly from unity (p >= 0.07) and therefore, the total number of infants who experienced health problems did not differ significantly between the two types of follow-up.
The only statistically significant difference found between TSFU and HVFU regarding the health problems experienced by infants was related to the weight of infants at site B. 29 mothers (15.7%) receiving TSFU versus 42 (24.9%) expressed concerns about their infants' weight (p< 0.05).
The rates of breastfeeding were similar for TSFU and HVFU at each one of the sites. The maternal confidence scores were also similar, and the differences in the infants' health problems reported by their mothers were not significantly different between the interventions. Therefore, both interventions appeared to obtain similar results according to the primary health outcomes considered in the clinical analysis. However, the authors stated that the study was not sufficiently powered to show equivalence of TSFU and HVFU.
Measure of benefits used in the economic analysis
No summary measure of benefits was used in the economic analysis and health outcomes were left disaggregated. The study was therefore categorised as a cost-consequences study.
Most of the resource quantities were reported separately from the costs. The direct costs considered in the study were those of the health service, and included: routine and unscheduled visits to family physicians or paediatricians (i.e. well baby care, consultations with paediatricians, home visits and follow-up visits), visits to other health professionals (i.e. lactation consultants, breastfeeding clinics, and chiropractors), emergency department visits, hospital admissions, laboratory tests, and PHN contacts. The costs of medications and other costs assumed by patients (i.e. transportation, meals, parking and child care) were also included. Resource utilisation was obtained from mothers' reports and was verified against medical records. The cost/resource boundary adopted for the analysis of direct costs appears to have been that of the third party payer. The sources of the direct costs were the Ontario Health Insurance fees and data obtained from the site A. Therefore, the estimation of unit costs was based on actual data. Discounting did not appear to have been performed, and it was not relevant since the period over which costs were incurred was shorter than 2 years. The study reported the total average costs per 100 infants. The price year was 1999. The authors reported that costs of no breastfeeding were not included in the analysis, but they did not provide any justification for this exclusion.
Statistical analysis of costs
No statistical analyses of costs were performed.
The indirect costs (i.e. those derived from the work loss of parents) were also considered in the economic analysis, although the authors did not provide a rationale for their inclusion. The perspective adopted in the study was that of society. The sources of the quantity data were telephone interviews to the mothers. The source of the unit costs used to value productivity losses was not reported, and therefore, it cannot be established whether the whole estimation of costs was based on actual data. Discounting was not performed and it was correctly not relevant since the period of study was shorter than 2 years. Resource quantities and costs were not reported separately. Moreover, the indirect costs were reported jointly with the direct costs borne by patients (except for medication costs). The price year was 1999.
No sensitivity analyses were performed.
Estimated benefits used in the economic analysis
The reader is referred to the 'Effectiveness Results' section reported earlier.
The total average costs per 100 children were: Can$18,747.75 at site A and Can$11,782.97 at site B for TSFU, versus Can$22,256.78 at site A and Can$26,419.56 at site B for HVFU.
Synthesis of costs and benefits
Not applicable due to the cost-consequences approach undertaken.
The conclusion reached by the authors was that HVFU and TSFU did not present significant differences in outcomes (although clinical equivalence was not demonstrated), but that the costs of HVFU were substantially higher than those of TSFU.
CRD COMMENTARY - Selection of comparators
None of the types of follow-up compared at analysis was specifically stated as comparator. HVFU was current practice in both hospitals included in the study, while TSFU appeared to have been current practice only at one of the sites. You, as the user of this database, should decide whether these are widely used types of postpartum follow-up in your own setting.
Validity of estimate of measure of effectiveness
The analysis was based on a randomised controlled trial, which was appropriate for the study question. The authors did not show evidence that the study sample was representative of the study population, although the facts that women from two different hospitals were included and that the sample size was large, may have increased the likelihood that the study sample was representative of the study population. The authors stated that regression analyses were used in order to control for confounding factors and reported the potential confounding factors. Blind assessment of the outcome was performed and the method of randomisation was reported. The overall internal validity of the analysis appears to have been high. However, there are some issues relating to the analysis. Although the authors stated that the basis for the effectiveness analysis was intention-to-treat, not all of the primary outcomes were reported following an intention-to-treat analysis: some were reported on the basis of available information at each point in time. Therefore, some caution is required when extrapolating the results to the overall study population. The nature of the intervention makes it difficult to assess whether the primary outcomes used in the effectiveness analysis could be a result of the interventions or of other factors. For example, breastfeeding may be more related to mothers' preferences rather than with the type of postpartum follow-up they receive. It would have been helpful to have considered other effectiveness outcomes, such as patients' satisfaction with the follow-up care provided.
Validity of estimate of measure of benefit
No summary measure of benefit was considered for the economic analysis. The study was therefore categorised as a cost-consequences analysis (see 'validity of effectiveness' comments earlier). The study could have been classified as a cost-minimisation analysis due to the lack of statistically significant difference in the outcome measures used in the effectiveness analysis, but the authors stated that the study did not have sufficient power to show the equivalence of TSFU and HVFU.
Validity of estimate of costs
The perspective adopted was a societal one and all the costs relevant to this perspective appear to have been included. The costs of not breastfeeding were omitted, but their inclusion is unlikely to affect the authors' conclusions. Most of the resource quantities and costs were reported separately and the price year was given, which may enhance reflation exercises to other settings. However, the indirect costs were reported jointly with the costs borne by patients, which would make reflation exercises difficult for such cost categories. No statistical analyses of resource quantities and costs were performed. Costs were specific to the study setting.
The authors did not make appropriate comparisons of their findings with those from other studies. The authors reported that the generalisability of the results is limited due to the facts that both the composition of the sub-samples and the implementation of the protocol differed at the two sites. They stated that the results should only be generalised to low-risk mothers and infants discharged together within two days of delivery and who are fluent in the main language of the community in which they reside.
Implications of the study
The authors suggested that it might not be necessary to provide a postpartum home visit to every mother, but that a standardised telephone call from a PHN could be used to identify the need for further intervention. They also stated that further research is needed to determine whether TSFU and HVFU result in the same outcomes, since the study had enough power to detect clinically significant differences but it did not have power to show equivalence of TSFU and HVFU.
Source of funding
Supported by The Physicians' Services Incorporated Foundation
Steel O'Connor K O, Mowat D L, Scott H M, Carr P A, Dorland J L, Young Tai K F W. A randomized trial of two public health nurse follow-up programs after early obstetrical discharge. An examination of breastfeeding rates, maternal confidence and utilization and costs of health services. Canadian Journal of Public Health 2003; 94(2): 98-103
Other publications of related interest
Carty E M, Bradley C F. A randomised controlled evaluation of early postpartum hospital discharge. Birth 1990;17:199-204.
Escobar G J, Braveman P A, Ackerson L, Odouli R, Coleman-Phox K, Capra A M, et al. A randomised comparison of home visits and hospital-based group follow-up visits after early postpartum discharge. Pediatrics 2001;108:719-27.
Gagnon A J, Dougherty G, Jimenez V, Leduc N. Randomised trial of postpartum care after hospital discharge. Pediatrics 2002;109:1074-80.
Lieu T A, Braveman P A, Escobar G J, Fischer A F, Jensvold N G, Capra A M. A randomised comparison of home and clinic follow-up visits after early postpartum hospital discharge. Pediatrics 2000;105:1058-65.
Morrell C J, Spiby H, Stewart P, Walters S, Morgan A. Cost and effectiveness of community postnatal support workers: Randomised controlled trial. BMJ 2000;321:593-98.
Subject indexing assigned by NLM
Adult; Breast Feeding; Comparative Study; Female; Follow-Up Studies; Health Care Costs; Health Services Research; House Calls; Humans; Mother-Child Relations; Ontario; Patient Discharge; Postnatal Care /organization & Public Health nursing /organization & Research Support, Non-U.S. Gov't; Telephone; administration; administration
Date bibliographic record published
Date abstract record published